| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.030090270812437, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05015045135406219, | |
| "grad_norm": 7.437633514404297, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.9756, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10030090270812438, | |
| "grad_norm": 3.9726901054382324, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.8322, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15045135406218657, | |
| "grad_norm": 3.629199743270874, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.5489, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.20060180541624875, | |
| "grad_norm": 3.2304935455322266, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.4407, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25075225677031093, | |
| "grad_norm": 3.2816317081451416, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.4424, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.30090270812437314, | |
| "grad_norm": 3.1936333179473877, | |
| "learning_rate": 3e-06, | |
| "loss": 0.4074, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3510531594784353, | |
| "grad_norm": 3.1005301475524902, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.4104, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4012036108324975, | |
| "grad_norm": 3.329033851623535, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.4049, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45135406218655966, | |
| "grad_norm": 2.9963624477386475, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.3773, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5015045135406219, | |
| "grad_norm": 2.877650260925293, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3659, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.551654964894684, | |
| "grad_norm": 3.1805803775787354, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.3686, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6018054162487463, | |
| "grad_norm": 2.518958806991577, | |
| "learning_rate": 6e-06, | |
| "loss": 0.362, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6519558676028084, | |
| "grad_norm": 2.588172674179077, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.3603, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7021063189568706, | |
| "grad_norm": 3.022862672805786, | |
| "learning_rate": 7e-06, | |
| "loss": 0.3835, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7522567703109327, | |
| "grad_norm": 2.3474695682525635, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3567, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.802407221664995, | |
| "grad_norm": 2.847740411758423, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3566, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8525576730190572, | |
| "grad_norm": 2.8755927085876465, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3552, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9027081243731193, | |
| "grad_norm": 2.488140106201172, | |
| "learning_rate": 9e-06, | |
| "loss": 0.3574, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9528585757271816, | |
| "grad_norm": 3.7667624950408936, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.34, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0030090270812437, | |
| "grad_norm": 2.2906906604766846, | |
| "learning_rate": 1e-05, | |
| "loss": 0.332, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.053159478435306, | |
| "grad_norm": 2.970755100250244, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.245, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.103309929789368, | |
| "grad_norm": 2.735118865966797, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.2632, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.1534603811434303, | |
| "grad_norm": 2.1645102500915527, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.2485, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.2036108324974926, | |
| "grad_norm": 2.3044114112854004, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.2545, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.2537612838515546, | |
| "grad_norm": 2.445683240890503, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.2398, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.3039117352056169, | |
| "grad_norm": 1.9902832508087158, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.2374, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.3540621865596791, | |
| "grad_norm": 2.571229934692383, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.2532, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.4042126379137412, | |
| "grad_norm": 2.24031662940979, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.2466, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.4543630892678034, | |
| "grad_norm": 2.2203283309936523, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.2414, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.5045135406218657, | |
| "grad_norm": 2.821672201156616, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.2612, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.5546639919759278, | |
| "grad_norm": 2.279815196990967, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.2361, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.60481444332999, | |
| "grad_norm": 2.2528128623962402, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.2337, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.6549648946840523, | |
| "grad_norm": 2.5302581787109375, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.2407, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.7051153460381143, | |
| "grad_norm": 2.0469911098480225, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.2603, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.7552657973921764, | |
| "grad_norm": 2.5774879455566406, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.2441, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.8054162487462388, | |
| "grad_norm": 2.362243175506592, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.2315, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.855566700100301, | |
| "grad_norm": 2.5620100498199463, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.2388, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.905717151454363, | |
| "grad_norm": 2.1473169326782227, | |
| "learning_rate": 9e-06, | |
| "loss": 0.23, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.9558676028084254, | |
| "grad_norm": 2.693861722946167, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.2491, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.0060180541624875, | |
| "grad_norm": 1.6720950603485107, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.2144, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0060180541624875, | |
| "eval_loss": 0.3075518310070038, | |
| "eval_runtime": 4546.1548, | |
| "eval_samples_per_second": 2.367, | |
| "eval_steps_per_second": 0.148, | |
| "eval_wer": 0.2114135961824298, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0561685055165495, | |
| "grad_norm": 2.0117599964141846, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.1061, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.106318956870612, | |
| "grad_norm": 1.9226505756378174, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.1171, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.156469408224674, | |
| "grad_norm": 2.096928358078003, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.1135, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.206619859578736, | |
| "grad_norm": 1.7984790802001953, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.1164, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.2567703109327986, | |
| "grad_norm": 1.6253271102905273, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.1136, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.3069207622868606, | |
| "grad_norm": 1.7676957845687866, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.1207, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.3570712136409226, | |
| "grad_norm": 2.040688991546631, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.1148, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.407221664994985, | |
| "grad_norm": 2.272294282913208, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.1222, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.457372116349047, | |
| "grad_norm": 2.0784871578216553, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.1188, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.507522567703109, | |
| "grad_norm": 1.869183897972107, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.1187, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.5576730190571713, | |
| "grad_norm": 2.0832390785217285, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.1176, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.6078234704112337, | |
| "grad_norm": 1.9388813972473145, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.1146, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.657973921765296, | |
| "grad_norm": 1.659645676612854, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.116, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.7081243731193583, | |
| "grad_norm": 1.7133307456970215, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.1201, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.7582748244734203, | |
| "grad_norm": 2.0739212036132812, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.1177, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.8084252758274824, | |
| "grad_norm": 1.760362148284912, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.1071, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.8585757271815444, | |
| "grad_norm": 1.979827880859375, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.1141, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.908726178535607, | |
| "grad_norm": 1.8925753831863403, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.1272, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.958876629889669, | |
| "grad_norm": 1.9880926609039307, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.1189, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.0090270812437314, | |
| "grad_norm": 1.2164592742919922, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.1091, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0591775325977935, | |
| "grad_norm": 1.2302759885787964, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.0472, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.1093279839518555, | |
| "grad_norm": 1.2639251947402954, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.048, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.1594784353059175, | |
| "grad_norm": 1.2657603025436401, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.0488, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.20962888665998, | |
| "grad_norm": 1.3162034749984741, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.0486, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.259779338014042, | |
| "grad_norm": 1.4205294847488403, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0484, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.3099297893681046, | |
| "grad_norm": 1.3729196786880493, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.0512, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.3600802407221666, | |
| "grad_norm": 1.333731770515442, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.05, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.4102306920762286, | |
| "grad_norm": 1.2979406118392944, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.0535, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.4603811434302907, | |
| "grad_norm": 1.1790673732757568, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.0598, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.510531594784353, | |
| "grad_norm": 1.7529319524765015, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0539, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.560682046138415, | |
| "grad_norm": 1.5352625846862793, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.0494, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.6108324974924777, | |
| "grad_norm": 1.707581639289856, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.051, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.6609829488465397, | |
| "grad_norm": 1.3295681476593018, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.0545, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.711133400200602, | |
| "grad_norm": 1.3010529279708862, | |
| "learning_rate": 7e-06, | |
| "loss": 0.0534, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.761283851554664, | |
| "grad_norm": 1.0611276626586914, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.0509, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.8114343029087263, | |
| "grad_norm": 1.1071949005126953, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.0496, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.8615847542627884, | |
| "grad_norm": 1.507038950920105, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.0494, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.9117352056168504, | |
| "grad_norm": 1.2737189531326294, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.052, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.961885656970913, | |
| "grad_norm": 1.8823206424713135, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.053, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.012036108324975, | |
| "grad_norm": 0.7559254765510559, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0472, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.012036108324975, | |
| "eval_loss": 0.33296459913253784, | |
| "eval_runtime": 4527.5772, | |
| "eval_samples_per_second": 2.377, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.20142963310646952, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.062186559679037, | |
| "grad_norm": 0.8223534822463989, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.0218, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.112337011033099, | |
| "grad_norm": 0.7893447279930115, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0201, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.162487462387162, | |
| "grad_norm": 0.657762348651886, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0203, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.212637913741224, | |
| "grad_norm": 0.9496999979019165, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.018, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.262788365095286, | |
| "grad_norm": 1.3166756629943848, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0203, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.312938816449348, | |
| "grad_norm": 0.8114544749259949, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.022, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.36308926780341, | |
| "grad_norm": 1.0165122747421265, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.0188, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.413239719157472, | |
| "grad_norm": 0.8669913411140442, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.0204, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.463390170511534, | |
| "grad_norm": 1.483972430229187, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.0199, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.513540621865597, | |
| "grad_norm": 1.1734397411346436, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0194, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.563691073219659, | |
| "grad_norm": 0.5651412606239319, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.0219, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.613841524573721, | |
| "grad_norm": 0.9663107395172119, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0215, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.663991975927783, | |
| "grad_norm": 0.6868607401847839, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.0223, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.714142427281845, | |
| "grad_norm": 1.138351321220398, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.0215, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.764292878635907, | |
| "grad_norm": 0.7841187119483948, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0207, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.81444332998997, | |
| "grad_norm": 0.6683496236801147, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.0201, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.864593781344032, | |
| "grad_norm": 1.2674401998519897, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.0196, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.914744232698094, | |
| "grad_norm": 0.8776654601097107, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.0214, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.964894684052156, | |
| "grad_norm": 0.8897686004638672, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0201, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.015045135406218, | |
| "grad_norm": 0.6467758417129517, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.0176, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.0651955867602805, | |
| "grad_norm": 0.8919677734375, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.0099, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 5.115346038114343, | |
| "grad_norm": 0.23880383372306824, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.0089, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.1654964894684054, | |
| "grad_norm": 0.7303509712219238, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.0093, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.2156469408224675, | |
| "grad_norm": 0.8609201312065125, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.0083, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.2657973921765295, | |
| "grad_norm": 0.4490683078765869, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.009, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.315947843530592, | |
| "grad_norm": 0.5905369520187378, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.009, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.366098294884654, | |
| "grad_norm": 0.18808700144290924, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.0075, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.4162487462387165, | |
| "grad_norm": 0.8524003624916077, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0084, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.466399197592779, | |
| "grad_norm": 0.31983518600463867, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.0081, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.516549648946841, | |
| "grad_norm": 0.40100687742233276, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0077, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.566700100300903, | |
| "grad_norm": 1.5288898944854736, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.0077, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.616850551654965, | |
| "grad_norm": 0.15371347963809967, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0068, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.667001003009027, | |
| "grad_norm": 0.547875702381134, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.0078, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.717151454363089, | |
| "grad_norm": 0.3743741512298584, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.0105, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.767301905717152, | |
| "grad_norm": 1.20223867893219, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.0084, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.817452357071214, | |
| "grad_norm": 0.4010525047779083, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0074, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.867602808425276, | |
| "grad_norm": 0.2383079081773758, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.0097, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.917753259779338, | |
| "grad_norm": 0.9694074988365173, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.0082, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.9679037111334, | |
| "grad_norm": 0.4873412549495697, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.0094, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 6.018054162487463, | |
| "grad_norm": 0.35399332642555237, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0074, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.018054162487463, | |
| "eval_loss": 0.3955759108066559, | |
| "eval_runtime": 4515.9024, | |
| "eval_samples_per_second": 2.383, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.19721505123992802, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.068204613841525, | |
| "grad_norm": 0.32993289828300476, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.005, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 6.118355065195587, | |
| "grad_norm": 2.4931156635284424, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.0047, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.168505516549649, | |
| "grad_norm": 0.09684254229068756, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.0039, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 6.218655967903711, | |
| "grad_norm": 0.1377803087234497, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0041, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.268806419257773, | |
| "grad_norm": 0.12329312413930893, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0036, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 6.318956870611835, | |
| "grad_norm": 0.39106011390686035, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.0038, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.369107321965898, | |
| "grad_norm": 0.21228019893169403, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.0034, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.41925777331996, | |
| "grad_norm": 0.13951948285102844, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0035, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.469408224674022, | |
| "grad_norm": 0.15315498411655426, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.0045, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.519558676028084, | |
| "grad_norm": 0.09918154031038284, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0037, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.569709127382146, | |
| "grad_norm": 0.128856360912323, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.0034, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.619859578736209, | |
| "grad_norm": 0.35564741492271423, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.0035, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.670010030090271, | |
| "grad_norm": 0.09503413736820221, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.0047, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.720160481444333, | |
| "grad_norm": 0.2272273600101471, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.004, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.770310932798395, | |
| "grad_norm": 0.31332728266716003, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0038, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 6.820461384152457, | |
| "grad_norm": 0.20381228625774384, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0044, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.870611835506519, | |
| "grad_norm": 0.38565367460250854, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0052, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 6.920762286860581, | |
| "grad_norm": 0.1613508015871048, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.0037, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.970912738214644, | |
| "grad_norm": 0.23183399438858032, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.0042, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 7.021063189568706, | |
| "grad_norm": 0.09113147854804993, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0028, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.071213640922768, | |
| "grad_norm": 0.12553289532661438, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0026, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 7.12136409227683, | |
| "grad_norm": 0.0897342711687088, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.002, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 7.1715145436308925, | |
| "grad_norm": 0.07827088236808777, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0024, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 7.2216649949849545, | |
| "grad_norm": 0.055039647966623306, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0024, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.271815446339017, | |
| "grad_norm": 0.06994307041168213, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0021, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 7.3219658976930795, | |
| "grad_norm": 0.10969334840774536, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0023, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.3721163490471415, | |
| "grad_norm": 0.08186643570661545, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0022, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 7.422266800401204, | |
| "grad_norm": 0.08872943371534348, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0027, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.472417251755266, | |
| "grad_norm": 0.20957154035568237, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.0027, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.522567703109328, | |
| "grad_norm": 0.09654885530471802, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0021, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.57271815446339, | |
| "grad_norm": 0.10487361997365952, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0041, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.622868605817453, | |
| "grad_norm": 0.06899241358041763, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0025, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.673019057171515, | |
| "grad_norm": 0.06502360850572586, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0025, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 7.723169508525577, | |
| "grad_norm": 0.05213673412799835, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0026, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.773319959879639, | |
| "grad_norm": 0.11386407911777496, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0023, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 7.823470411233701, | |
| "grad_norm": 0.08614542335271835, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.002, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.873620862587764, | |
| "grad_norm": 0.12035714089870453, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0021, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 7.923771313941826, | |
| "grad_norm": 0.0683644562959671, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0023, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 7.973921765295888, | |
| "grad_norm": 0.059945497661828995, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.002, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 8.02407221664995, | |
| "grad_norm": 0.050406791269779205, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0019, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.02407221664995, | |
| "eval_loss": 0.4262579679489136, | |
| "eval_runtime": 4522.1238, | |
| "eval_samples_per_second": 2.38, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.1952691074082766, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.074222668004012, | |
| "grad_norm": 0.04868149757385254, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0016, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 8.124373119358074, | |
| "grad_norm": 0.0494823083281517, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0018, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 8.174523570712136, | |
| "grad_norm": 0.05864572152495384, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0017, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 8.224674022066198, | |
| "grad_norm": 0.05070830136537552, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0016, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 8.27482447342026, | |
| "grad_norm": 0.06109990179538727, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0017, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 8.324974924774324, | |
| "grad_norm": 0.045594800263643265, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0017, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 8.375125376128386, | |
| "grad_norm": 0.049554213881492615, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0021, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 8.425275827482448, | |
| "grad_norm": 0.11533299088478088, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0015, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 8.47542627883651, | |
| "grad_norm": 0.043509747833013535, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0018, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 8.525576730190572, | |
| "grad_norm": 0.14743666350841522, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.002, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 8.575727181544634, | |
| "grad_norm": 0.08023034036159515, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0017, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 8.625877632898696, | |
| "grad_norm": 0.1176542118191719, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0019, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.676028084252758, | |
| "grad_norm": 0.05018528178334236, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0018, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 8.72617853560682, | |
| "grad_norm": 0.051610466092824936, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0016, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 8.776328986960882, | |
| "grad_norm": 0.04370126128196716, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0016, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 8.826479438314944, | |
| "grad_norm": 0.06507682055234909, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0015, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.876629889669006, | |
| "grad_norm": 0.06819342076778412, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0019, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 8.926780341023068, | |
| "grad_norm": 0.06309027969837189, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0017, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 8.976930792377132, | |
| "grad_norm": 0.08104189485311508, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0015, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 9.027081243731194, | |
| "grad_norm": 0.04061448946595192, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0015, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 9.077231695085256, | |
| "grad_norm": 0.04562969505786896, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0018, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 9.127382146439318, | |
| "grad_norm": 0.050402842462062836, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0015, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 9.17753259779338, | |
| "grad_norm": 0.5361291170120239, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0019, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 9.227683049147442, | |
| "grad_norm": 0.0427754744887352, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0015, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 9.277833500501504, | |
| "grad_norm": 0.04894596338272095, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0015, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 9.327983951855567, | |
| "grad_norm": 0.04485371708869934, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0014, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 9.378134403209629, | |
| "grad_norm": 0.04477512091398239, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0015, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 9.42828485456369, | |
| "grad_norm": 0.04668520390987396, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0015, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 9.478435305917753, | |
| "grad_norm": 0.05096197500824928, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0017, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 9.528585757271815, | |
| "grad_norm": 0.04469752684235573, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0015, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 9.578736208625877, | |
| "grad_norm": 0.050533875823020935, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0014, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 9.62888665997994, | |
| "grad_norm": 0.04058687761425972, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0015, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 9.679037111334003, | |
| "grad_norm": 0.05510278418660164, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0014, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 9.729187562688065, | |
| "grad_norm": 0.05432584136724472, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0013, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 9.779338014042127, | |
| "grad_norm": 0.08223077654838562, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0014, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 9.829488465396189, | |
| "grad_norm": 0.051733825355768204, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0014, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.87963891675025, | |
| "grad_norm": 0.03811487555503845, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0013, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 9.929789368104313, | |
| "grad_norm": 0.03971032425761223, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0013, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.979939819458375, | |
| "grad_norm": 0.0468224436044693, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0016, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 10.030090270812437, | |
| "grad_norm": 0.04448791965842247, | |
| "learning_rate": 0.0, | |
| "loss": 0.0014, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.030090270812437, | |
| "eval_loss": 0.4447694718837738, | |
| "eval_runtime": 4528.3909, | |
| "eval_samples_per_second": 2.377, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.1960611749980443, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.030090270812437, | |
| "step": 5000, | |
| "total_flos": 5.43090104303616e+20, | |
| "train_loss": 0.0887994671344757, | |
| "train_runtime": 50408.8653, | |
| "train_samples_per_second": 3.174, | |
| "train_steps_per_second": 0.099 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 11, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.43090104303616e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |